Levenberg–Marquardt algorithm: Difference between revisions

Content deleted Content added
Alanb (talk | contribs)
Alanb (talk | contribs)
Line 41:
 
Obviously the absolute values of any choice depends on how well-scaled the initial problem is. Marquadt recommended starting with a value &lambda;<sub>0</sub> and a factor &nu;>1. Initiall set &lambda;=&lambda;<sub>0</sub>; and computing the residual sum of squares ''S''('''p''') after one step from the starting point with the damping factor of
&lambda;=&lambda;<sub>0</sub>; and secondly with &lambda;/&nu; If both of these are worse than the initial point then the damping is increased by successive multiplication by &nu; until a better point is found with a new damping factor of &lambda; &nu;<sup>k</sup>
for some ''k''.